Agent-BRACE improves LLM agent performance on long-horizon partially observable tasks by 5.3-14.5% through a decoupled belief state of verbalized atomic claims with certainty labels that keeps context length constant.
citation dossier · super hub
SWE-bench: Can Language Models Resolve Real-World GitHub Issues?
Language models have outpaced our ability to evaluate them effectively, but for their future development it is essential to study the frontier of their capabilities. We find real-world software engineering to be a rich, sustainable, and challenging testbed for evaluating the next generation of language models. To this end, we introduce SWE-bench, an evaluation framework consisting of $2,294$ software engineering problems drawn from real GitHub issues and corresponding pull requests across $12$ popular Python repositories. Given a codebase along with a description of an issue to be resolved, a language model is tasked with editing the codebase to address the issue. Resolving issues in SWE-bench frequently requires understanding and coordinating changes across multiple functions, classes, and even files simultaneously, calling for models to interact with execution environments, process extremely long contexts and perform complex reasoning that goes far beyond traditional code generation tasks. Our evaluations show that both state-of-the-art proprietary models and our fine-tuned model SWE-Llama can resolve only the simplest issues. The best-performing model, Claude 2, is able to solve a mere $1.96$% of the issues. Advances on SWE-bench represent steps towards LMs that are more practical, intelligent, and autonomous.
143Pith papers citing it
148reference links
cs.SEtop field · 49 papers
UNVERDICTEDtop verdict bucket · 124 papers
why this work matters in Pith
Pith tracks SWE-bench: Can Language Models Resolve Real-World GitHub Issues? because it crossed a citation-hub threshold.
hub tools
claims ledger
- abstract Language models have outpaced our ability to evaluate them effectively, but for their future development it is essential to study the frontier of their capabilities. We find real-world software engineering to be a rich, sustainable, and challenging testbed for evaluating the next generation of language models. To this end, we introduce SWE-bench, an evaluation framework consisting of $2,294$ software engineering problems drawn from real GitHub issues and corresponding pull requests across $12$ popular Python repositories. Given a codebase along with a description of an issue to be resolved, a
authors
co-cited works
representative citing papers
A new native-runtime benchmark reveals that current frontier AI agents succeed on at most 62 percent of realistic long-horizon CLI tasks.
PDEAgent-Bench is the first multi-metric, multi-library benchmark for AI-generated PDE solvers, evaluating executability, numerical accuracy, and efficiency across DOLFINx, Firedrake, and deal.II.
VibeServe demonstrates that AI agents can synthesize bespoke LLM serving systems end-to-end, remaining competitive with vLLM in standard settings while outperforming it in six non-standard scenarios involving unusual models, workloads, or hardware.
StabilizerBench is a new benchmark for evaluating AI agents on generating, optimizing, and making fault-tolerant stabilizer circuits for quantum error correction, with efficient verification and multi-tier scoring.
neuralCAD-Edit benchmark shows even the best foundation model (GPT 5.2) scores 53% lower than human CAD experts in acceptance trials for multimodal-instructed 3D model edits.
HWE-Bench is the first repository-level benchmark for LLM agents on real hardware bug repair, where the best agent fixes 70.7% of 417 tasks but drops below 65% on complex SoC projects.
The AI Scientist framework enables LLMs to independently conduct the full scientific process from idea generation to paper writing and review, demonstrated across three ML subfields with papers costing under $15 each.
OSWorld provides the first unified real-computer benchmark for open-ended multimodal agent tasks, exposing large performance gaps between humans and state-of-the-art LLM/VLM agents.
AEvo introduces a meta-agent that edits the evolution procedure or agent context based on accumulated state, outperforming baselines by 26% relative improvement on agentic benchmarks and achieving SOTA on open-ended tasks.
ActGuide-RL uses human action data as plan-style guidance in mixed-policy RL to overcome exploration barriers in LLM agents, matching SFT+RL performance on search benchmarks without cold-start training.
StepCodeReasoner aligns code reasoning with verifiable stepwise execution traces via print anchors and bi-level GRPO reinforcement learning, reaching SOTA results on CRUXEval (91.1%) and LiveCodeBench (86.5%) for a 7B model.
CppPerf-Mine produces CppPerf-DB, a benchmark of 347 real-world performance-improving C++ patches (39% multi-file) from 42 repositories to evaluate repository-level repair tools.
TacoMAS performs test-time co-evolution of agent capabilities and communication topology in LLM multi-agent systems via fast capability updates and slow meta-LLM topology edits, delivering 13.3% average gains over strong baselines on four benchmarks.
LLMs encode tool necessity in pre-generation hidden states at AUROC 0.89-0.96, enabling Probe&Prefill to reduce tool calls 48% with 1.7% accuracy loss, outperforming prompt and reasoning baselines.
SkillGuard extracts executable environment contracts from LLM skill documents to detect only relevant drifts, reporting zero false positives on 599 cases, 100% precision in known-drift tests, and raising one-round repair success from 10% to 78%.
AgentForesight trains a 7B model to perform online auditing of multi-agent LLM trajectories, detecting early decisive errors and outperforming larger models on custom and external benchmarks.
AgentEscapeBench shows LLM agents' success rates drop from 90% to 60% as tool-dependency depth increases from 5 to 25 steps, while humans drop only from 98% to 80%.
LLM agents exhibit persistent attack-selection biases as fixed traits independent of success rates, with a bias momentum effect that resists steering and yields no performance gain.
LLM agents reach only 50.6% accuracy on chemical cost estimation within 25% error even with tools, dropping with noise due to parsing, pack selection, and tool-use failures.
Switchcraft routes agentic tool-calling queries to the lowest-cost model that preserves correctness, reaching 82.9% accuracy and 84% cost reduction on five benchmarks.
MOSAIC-Bench demonstrates that nine production coding agents achieve 53-86% end-to-end attack success rates on staged innocuous tickets across 10 web substrates and 31 CWE classes, far higher than the 0-20.4% rates seen with direct prompts.
ProgramBench introduces 200 tasks where models must reconstruct full programs like FFmpeg or SQLite from docs alone; none of 9 evaluated LMs fully solve any task and the best passes 95% tests on only 3% of tasks while favoring monolithic code.
ARISE adds a data-flow-augmented repository graph and three-tier tool API to LLM agents, raising Function Recall@1 by 17 points, Line Recall@1 by 15 points, and Pass@1 repair rate to 22% on SWE-bench Lite.
citing papers explorer
-
Agent-BRACE: Decoupling Beliefs from Actions in Long-Horizon Tasks via Verbalized State Uncertainty
Agent-BRACE improves LLM agent performance on long-horizon partially observable tasks by 5.3-14.5% through a decoupled belief state of verbalized atomic claims with certainty labels that keeps context length constant.
-
WildClawBench: A Benchmark for Real-World, Long-Horizon Agent Evaluation
A new native-runtime benchmark reveals that current frontier AI agents succeed on at most 62 percent of realistic long-horizon CLI tasks.
-
PDEAgent-Bench: A Multi-Metric, Multi-Library Benchmark for PDE Solver Generation
PDEAgent-Bench is the first multi-metric, multi-library benchmark for AI-generated PDE solvers, evaluating executability, numerical accuracy, and efficiency across DOLFINx, Firedrake, and deal.II.
-
VibeServe: Can AI Agents Build Bespoke LLM Serving Systems?
VibeServe demonstrates that AI agents can synthesize bespoke LLM serving systems end-to-end, remaining competitive with vLLM in standard settings while outperforming it in six non-standard scenarios involving unusual models, workloads, or hardware.
-
StabilizerBench: A Benchmark for AI-Assisted Quantum Error Correction Circuit Synthesis
StabilizerBench is a new benchmark for evaluating AI agents on generating, optimizing, and making fault-tolerant stabilizer circuits for quantum error correction, with efficient verification and multi-tier scoring.
-
neuralCAD-Edit: An Expert Benchmark for Multimodal-Instructed 3D CAD Model Editing
neuralCAD-Edit benchmark shows even the best foundation model (GPT 5.2) scores 53% lower than human CAD experts in acceptance trials for multimodal-instructed 3D model edits.
-
HWE-Bench: Benchmarking LLM Agents on Real-World Hardware Bug Repair Tasks
HWE-Bench is the first repository-level benchmark for LLM agents on real hardware bug repair, where the best agent fixes 70.7% of 417 tasks but drops below 65% on complex SoC projects.
-
The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery
The AI Scientist framework enables LLMs to independently conduct the full scientific process from idea generation to paper writing and review, demonstrated across three ML subfields with papers costing under $15 each.
-
OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments
OSWorld provides the first unified real-computer benchmark for open-ended multimodal agent tasks, exposing large performance gaps between humans and state-of-the-art LLM/VLM agents.
-
Harnessing Agentic Evolution
AEvo introduces a meta-agent that edits the evolution procedure or agent context based on accumulated state, outperforming baselines by 26% relative improvement on agentic benchmarks and achieving SOTA on open-ended tasks.
-
Learning Agentic Policy from Action Guidance
ActGuide-RL uses human action data as plan-style guidance in mixed-policy RL to overcome exploration barriers in LLM agents, matching SFT+RL performance on search benchmarks without cold-start training.
-
StepCodeReasoner: Aligning Code Reasoning with Stepwise Execution Traces via Reinforcement Learning
StepCodeReasoner aligns code reasoning with verifiable stepwise execution traces via print anchors and bi-level GRPO reinforcement learning, reaching SOTA results on CRUXEval (91.1%) and LiveCodeBench (86.5%) for a 7B model.
-
CppPerf: An Automated Pipeline and Dataset for Performance-Improving C++ Commits
CppPerf-Mine produces CppPerf-DB, a benchmark of 347 real-world performance-improving C++ patches (39% multi-file) from 42 repositories to evaluate repository-level repair tools.
-
TacoMAS: Test-Time Co-Evolution of Topology and Capability in LLM-based Multi-Agent Systems
TacoMAS performs test-time co-evolution of agent capabilities and communication topology in LLM multi-agent systems via fast capability updates and slow meta-LLM topology edits, delivering 13.3% average gains over strong baselines on four benchmarks.
-
LLM Agents Already Know When to Call Tools -- Even Without Reasoning
LLMs encode tool necessity in pre-generation hidden states at AUROC 0.89-0.96, enabling Probe&Prefill to reduce tool calls 48% with 1.7% accuracy loss, outperforming prompt and reasoning baselines.
-
Skill Drift Is Contract Violation: Proactive Maintenance for LLM Agent Skill Libraries
SkillGuard extracts executable environment contracts from LLM skill documents to detect only relevant drifts, reporting zero false positives on 599 cases, 100% precision in known-drift tests, and raising one-round repair success from 10% to 78%.
-
AgentForesight: Online Auditing for Early Failure Prediction in Multi-Agent Systems
AgentForesight trains a 7B model to perform online auditing of multi-agent LLM trajectories, detecting early decisive errors and outperforming larger models on custom and external benchmarks.
-
AgentEscapeBench: Evaluating Out-of-Domain Tool-Grounded Reasoning in LLM Agents
AgentEscapeBench shows LLM agents' success rates drop from 90% to 60% as tool-dependency depth increases from 5 to 25 steps, while humans drop only from 98% to 80%.
-
CyBiasBench: Benchmarking Bias in LLM Agents for Cyber-Attack Scenarios
LLM agents exhibit persistent attack-selection biases as fixed traits independent of success rates, with a bias momentum effect that resists steering and yields no performance gain.
-
Can Agents Price a Reaction? Evaluating LLMs on Chemical Cost Reasoning
LLM agents reach only 50.6% accuracy on chemical cost estimation within 25% error even with tools, dropping with noise due to parsing, pack selection, and tool-use failures.
-
Switchcraft: AI Model Router for Agentic Tool Calling
Switchcraft routes agentic tool-calling queries to the lowest-cost model that preserves correctness, reaching 82.9% accuracy and 84% cost reduction on five benchmarks.
-
MOSAIC-Bench: Measuring Compositional Vulnerability Induction in Coding Agents
MOSAIC-Bench demonstrates that nine production coding agents achieve 53-86% end-to-end attack success rates on staged innocuous tickets across 10 web substrates and 31 CWE classes, far higher than the 0-20.4% rates seen with direct prompts.
-
ProgramBench: Can Language Models Rebuild Programs From Scratch?
ProgramBench introduces 200 tasks where models must reconstruct full programs like FFmpeg or SQLite from docs alone; none of 9 evaluated LMs fully solve any task and the best passes 95% tests on only 3% of tasks while favoring monolithic code.
-
ARISE: A Repository-level Graph Representation and Toolset for Agentic Fault Localization and Program Repair
ARISE adds a data-flow-augmented repository graph and three-tier tool API to LLM agents, raising Function Recall@1 by 17 points, Line Recall@1 by 15 points, and Pass@1 repair rate to 22% on SWE-bench Lite.
-
NeuroState-Bench: A Human-Calibrated Benchmark for Commitment Integrity in LLM Agent Profiles
NeuroState-Bench is a human-calibrated benchmark with 144 tasks and 306 side-query probes showing that commitment integrity in LLM agent profiles diverges from task success, with 31 of 32 profiles changing rank under integrity evaluation.
-
Reward Hacking Benchmark: Measuring Exploits in LLM Agents with Tool Use
The Reward Hacking Benchmark shows RL post-training raises exploit rates in tool-using LLM agents from 0.6% to 13.9%, with environmental hardening cutting exploits by 87.7% relative without lowering task success.
-
The Partial Testimony of Logs: Evaluation of Language Model Generation under Confounded Model Choice
An identification theorem shows that a randomized experiment and simulator together recover causal model values from confounded logs, with logs used only afterward to reduce estimation error.
-
CacheFlow: Efficient LLM Serving with 3D-Parallel KV Cache Restoration
CacheFlow cuts TTFT by 10-62% in batched LLM serving via 3D-parallel KV cache restoration and a two-pointer scheduler that overlaps recompute and I/O.
-
Empowering Autonomous Debugging Agents with Efficient Dynamic Analysis
ADI equips AI debugging agents with function-level interaction via a new execution trace structure, raising SWE-bench Verified resolution to 63.8% at $1.28 per task and delivering 6-18% gains when added to existing agents.
-
ClawMark: A Living-World Benchmark for Multi-Turn, Multi-Day, Multimodal Coworker Agents
ClawMark is a new benchmark for multi-turn multi-day multimodal coworker agents in stateful evolving services, with deterministic Python checkers showing frontier models achieve only 20% strict task success.
-
Assessing the Impact of Requirement Ambiguity on LLM-based Function-Level Code Generation
Orchid benchmark shows requirement ambiguity degrades LLM code generation performance across all models, with advanced models hit hardest, and LLMs rarely detect or resolve the ambiguity themselves.
-
Do Agents Dream of Root Shells? Partial-Credit Evaluation of LLM Agents in Capture the Flag Challenges
LLM agents reach only 35% average checkpoint completion on ten realistic CTF challenges in a new open benchmark with automated partial-credit scoring.
-
DebugRepair: Enhancing LLM-Based Automated Program Repair via Self-Directed Debugging
DebugRepair improves LLM-based automated program repair by adding test semantic purification, simulated instrumentation, and debugging-driven conversational repair, fixing 224 Defects4J bugs with GPT-3.5 (26.2% above prior SOTA) and 295 with DeepSeek-V3.
-
Choose Your Own Adventure: Non-Linear AI-Assisted Programming with EvoGraph
EvoGraph turns linear AI-assisted programming into a manipulable graph of branching histories, reducing cognitive load and enabling better iteration according to a user study with 20 developers.
-
Self-Consistency from Only Two Samples: CoT-PoT Ensembling for Efficient LLM Reasoning
CoT-PoT ensembling achieves self-consistency accuracy in LLMs with only two samples for 78.6% of tasks, reducing computation by 9.3x compared to standard methods.
-
SkillFlow:Benchmarking Lifelong Skill Discovery and Evolution for Autonomous Agents
SkillFlow benchmark shows lifelong skill evolution yields modest gains for some models like Claude Opus 4.6 but limited or negative utility for others despite high skill usage.
-
Feedback-Driven Execution for LLM-Based Binary Analysis
FORGE uses a reasoning-action-observation loop and Dynamic Forest of Agents to perform scalable LLM-based binary analysis, finding 1,274 vulnerabilities across 591 of 3,457 real-world firmware binaries at 72.3% precision and broader coverage than prior methods.
-
RealVuln: Benchmarking Rule-Based, General-Purpose LLM, and Security-Specialized Scanners on Real-World Code
RealVuln benchmark finds security-specialized scanners outperform general-purpose LLMs and rule-based SAST tools on hand-labeled vulnerable Python code under F3 scoring, with all artifacts released.
-
CodeComp: Structural KV Cache Compression for Agentic Coding
CodeComp uses Joern-extracted Code Property Graph priors for training-free structural KV cache compression, outperforming attention-only baselines on bug localization and code generation while matching full-context patch quality.
-
FinTrace: Holistic Trajectory-Level Evaluation of LLM Tool Calling for Long-Horizon Financial Tasks
FinTrace supplies trajectory-level metrics for LLM financial tool calling, exposing gaps in information use and output quality, while its preference dataset enables DPO training that boosts intermediate metrics.
-
Generate, Filter, Control, Replay: A Comprehensive Survey of Rollout Strategies for LLM Reinforcement Learning
This survey introduces the Generate-Filter-Control-Replay (GFCR) taxonomy to structure rollout pipelines for RL-based post-training of reasoning LLMs.
-
FrontierFinance: A Long-Horizon Computer-Use Benchmark of Real-World Financial Tasks
FrontierFinance benchmark shows human financial experts outperform state-of-the-art LLMs by achieving higher scores and more client-ready outputs on realistic long-horizon tasks.
-
Benchmarking and Evaluating VLMs for Software Architecture Diagram Understanding
SADU benchmark shows top VLMs reach only 70% accuracy on software architecture diagram tasks, revealing gaps in visual reasoning for engineering artifacts.
-
Toward Executable Repository-Level Code Generation via Environment Alignment
EnvGraph improves executable repository-level code generation by jointly modeling external dependencies and internal references through a dual-layer environment representation and targeted iterative alignment.
-
AgentHazard: A Benchmark for Evaluating Harmful Behavior in Computer-Use Agents
AgentHazard benchmark shows computer-use agents remain highly vulnerable, with attack success rates reaching 73.63% on models like Qwen3-Coder powering Claude Code.
-
Evaluating the Formal Reasoning Capabilities of Large Language Models through Chomsky Hierarchy
LLMs display clear performance stratification on formal language tasks aligned with Chomsky hierarchy complexity levels, limited by severe efficiency barriers rather than absolute capability.
-
MatClaw: An Autonomous Code-First LLM Agent for End-to-End Materials Exploration
MatClaw is a code-first LLM agent that autonomously executes end-to-end materials workflows by generating and running Python scripts on remote clusters, achieving reliable code generation via memory architecture and RAG while requiring guided interventions for tacit knowledge.
-
AgentSZZ: Teaching the LLM Agent to Play Detective with Bug-Inducing Commits
AgentSZZ is an LLM-agent framework that identifies bug-inducing commits with up to 27.2% higher F1 scores than prior methods by enabling adaptive exploration and causal tracing, especially for cross-file and ghost commits.
-
REAP: Automatic Curation of Coding Agent Benchmarks from Interactive Production Usage
REAP automatically curates production-derived benchmarks for AI coding agents via LLM classification and stability checks, producing the Harvest benchmark with model solve rates of 42.9-58.2%.
-
$\tau^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment
τ²-bench provides a Dec-POMDP-based telecom domain with compositional task generation and a tool-constrained user simulator to measure agent performance drops in dual-control versus single-control settings.